Glass capillary gas chromatography is used to separate numerous constituents of human physiological fluids. The generated multicomponent analytical data are subjected to computational pattern recognition analysis, using threshold logic units and feature extraction techniques. Preliminary results on diabetes mellitus indicate over 90% correct prediction, and several metabolites selected by the feature extraction are under study. Polyol profiles of human cerebrospinal fluids are determined to study etiology of neurological disorders. Some human studies are presently parallelled by investigations of suitable animal models.